222 BOTANICAL GAZETTE [SEPTEMBER 
embryo sac has been observed in a number of cases (COULTER and 
CHAMBERLAIN 3), the condition reaching its extreme in Loranthus 
(HOFMEISTER 6 and TREUB 14), in which scarcely more than the 
lower one-tenth of the sac becomes filled with permanent endo- 
sperm tissue. Among the Labiatae the work of TULASNE (15), 
HoFrMEIsTER (6), and VESQUE (16) shows this to be the prevailing 
condition in several genera. In Stachys sylvatica TULASNE figures 
endosperm developing in the antipodal region of a slightly curved 
sac, but without the presence of a special chamber; and in Beton- 
ica a condition which may well represent a later stage in the same 
situation. The figures of HOFMEISTER indicate that in Lamium 
the endosperm lobe is well developed before fertilization, as in 
Physostegia. Although no antipodals and only two “Keim- 
blischen”’ are represented, HOFMEISTER has figured stages which 
correspond approximately to those shown in figs. 7, 8, 14, and 16 
of this paper. 
In all of these cases the embryo is brought into contact with the 
endosperm by the great elongation of the micropylar cell of the 
proembryo. The earliest clearly observed stage in Physostegia is 
shown in fig. 13. Here the first division in the fertilized egg has 
occurred, and the micropylar cell by its great elongation is pushing 
the chalazal cell into the endosperm, the cells of which at this time 
are relatively few in number. Nearly all the elongation is accom- 
plished by the one cell, but this soon divides to several (figs. 14 
and 16). 
The first division in the chalazal cell is longitudinal (fig. 15), 
as is also the second. Each of the four resulting cells is then 
divided into two by a transverse wall (fig. 16), and the subsequent 
divisions proceed with much regularity (figs. 17 and 18). 
At the time when the embryo becomes imbedded in the endo- 
sperm, the micropylar and endosperm lobes are approximately 
equal in size. The former, as has been noted above, disorganizes 
and in later stages becomes entirely obliterated, while the latter 
increases rapidly in size owing to the active growth of the endo- 
sperm. This growth is accomplished at the expense of the cells of 
the integument, which in the mature seed is recognizable as only 
one or two layers of cells (figs. 19 and 20). At the same time the 
